<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Faith  Innocent on Medium]]></title>
        <description><![CDATA[Stories by Faith  Innocent on Medium]]></description>
        <link>https://medium.com/@faithinnocent774?source=rss-5fdc0e3bfa43------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*eRNJW39RNMzgn1EdjtQLyg.jpeg</url>
            <title>Stories by Faith  Innocent on Medium</title>
            <link>https://medium.com/@faithinnocent774?source=rss-5fdc0e3bfa43------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 12:12:21 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@faithinnocent774/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[What I Learned in Week 11: Data Structures and Algorithms — Practical Lessons, Challenges, and…]]></title>
            <link>https://medium.com/@faithinnocent774/what-i-learned-in-week-11-data-structures-and-algorithms-practical-lessons-challenges-and-f66bfbbf2c1c?source=rss-5fdc0e3bfa43------2</link>
            <guid isPermaLink="false">https://medium.com/p/f66bfbbf2c1c</guid>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[data-structures]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[algorithms]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Faith  Innocent]]></dc:creator>
            <pubDate>Mon, 16 Nov 2026 11:36:32 GMT</pubDate>
            <atom:updated>2026-11-16T11:36:32.080Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>What I Learned in Week 11: Data Structures and Algorithms — Practical Lessons, Challenges, and Growth</strong></h3><h3>What I Learned in Week 11: Data Structures and Algorithms</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VMGUk-OGgo1xPTqIrKyxUw.jpeg" /><figcaption>Created by Faith Innocent</figcaption></figure><p>This week, I learned how to use data structures and algorithms to solve practical problems more efficiently. One of my biggest takeaways was that getting the correct answer is important, but finding an efficient way to get that answer is equally important.</p><p>During Week 11, I practised arrays, two pointers, hashing, linked lists, stacks, recursion, and binary search. I also explored how some of these concepts can be applied to SQL queries and real-world data problems.</p><h3>1. Working with Arrays and Two Pointers</h3><p>I learned how the sliding window technique can help solve problems involving consecutive elements in an array.</p><p>For example, when finding the largest total of a given number of consecutive sales days, instead of calculating every possible group from scratch, I can reuse the previous window’s sum. This makes the solution faster.</p><p>I also learned the two-pointer technique. It uses two positions in a sorted array and moves them towards each other to find a pair of values that adds up to a target. This can solve the problem in O(n) time.</p><h3>2. Understanding Hashing</h3><p>Hashing was another important topic I studied.</p><p>I used dictionaries and sets to solve problems such as grouping anagrams and finding the longest sequence of consecutive numbers.</p><p>I discovered that sets are useful because they allow efficient membership checks on average. This can significantly reduce the time needed to solve certain problems.</p><p>One important lesson was that a solution can contain a loop inside another loop and still have O(n) time complexity if each element is processed only a limited number of times overall.</p><h3>3. Linked Lists and Stacks</h3><p>I practised reversing a linked list by changing the direction of its node references.</p><p>I also learned how stacks work using the Last In, First Out (LIFO) principle.</p><p>One practical example was checking whether brackets in a string are properly balanced. A stack helps keep track of opening brackets and ensures that each closing bracket matches the correct opening bracket.</p><p>I also used a monotonic stack to solve the daily temperatures problem. The goal was to find how many days each temperature had to wait before a warmer day appeared.</p><p>This helped me understand how a stack can make a problem more efficient by avoiding repeated comparisons.</p><h3>4. Learning Recursion</h3><p>Recursion was another interesting concept.</p><p>I used recursion to flatten nested lists into a single list while preserving the original order of the elements.</p><p>I learned that recursion requires two important parts:</p><ul><li><strong>Base case:</strong> The condition that stops the recursive calls.</li><li><strong>Recursive case:</strong> The part where the function calls itself to solve a smaller version of the problem.</li></ul><p>I also learned that deeply nested structures can exceed Python’s recursion limit, and that an explicit stack can be used as an alternative.</p><h3>5. Searching and Binary Search</h3><p>I learned how binary search works by repeatedly dividing a sorted search space into smaller sections.</p><p>I applied it to find the first and last positions of a target value in a sorted list.</p><p>I also used binary search to solve a shipping capacity problem. The task was to find the smallest truck capacity that could deliver all parcels within a given number of days while keeping the parcels in their original order.</p><p>The important idea was that if a particular truck capacity works, any larger capacity will also work. This property makes binary search suitable for finding the smallest valid answer.</p><h3>6. Applying Data Structures and Algorithms to SQL</h3><p>I also worked on four SQL practice tasks that connected database operations to the concepts I learned.</p><p>These included:</p><ul><li>Using GROUP BY to organise and count records.</li><li>Using NOT EXISTS to identify records without matching entries.</li><li>Using window functions and prefix sums to calculate running totals.</li><li>Using sliding-window calculations to identify the warmest seven-day period for different cities.</li></ul><p>This helped me understand that algorithmic thinking is useful not only when writing Python code but also when working with databases.</p><h3>7. Challenges I Faced and How I Addressed Them</h3><p>One of my biggest challenges was identifying the right technique for each problem.</p><p>Sometimes, a simple solution using nested loops could produce the correct answer, but it might become too slow when working with large amounts of data.</p><p>I learned to look for better approaches by using sets, dictionaries, stacks, two pointers, and binary search.</p><p>I also paid attention to edge cases, including empty lists, repeated values, invalid inputs, unmatched brackets, and values that do not exist in a sorted list.</p><p>Testing these situations helped me understand why reviewing a solution is just as important as writing it.</p><h3>8. What I Would Do Differently Next Time</h3><p>In future projects, I want to spend more time planning before writing code.</p><p>I would first understand the problem, identify possible edge cases, choose the most suitable data structure, and estimate the time and space complexity.</p><p>I would also test my solutions with small examples before running larger tests.</p><p>For SQL tasks, I would make sure to verify table names, column names, and database schemas before executing queries.</p><h3>Conclusion: My Biggest Takeaway</h3><p>Week 11 taught me that data structures and algorithms are not just theoretical programming concepts. They are practical tools that help us organise information, solve problems, and improve performance.</p><p>I now have a better understanding of how to choose the right technique for a problem and how to think about the efficiency of my solutions.</p><p>The most important lesson I am taking forward is that writing code that works is only the beginning. Understanding why it works, how efficiently it runs, and how it can be improved is what makes me a better problem solver.</p><p>I look forward to applying these concepts in future projects and continuing to improve my programming skills.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f66bfbbf2c1c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Week 8 at DataraFlow: Turning Trade Data into Meaningful Insights]]></title>
            <link>https://medium.com/@faithinnocent774/week-8-at-dataraflow-turning-trade-data-into-meaningful-insights-5ce1f3e8fa00?source=rss-5fdc0e3bfa43------2</link>
            <guid isPermaLink="false">https://medium.com/p/5ce1f3e8fa00</guid>
            <category><![CDATA[python]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Faith  Innocent]]></dc:creator>
            <pubDate>Fri, 30 Oct 2026 07:49:54 GMT</pubDate>
            <atom:updated>2026-10-30T07:49:54.857Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eDO_XzFJXXxolGlPsEs2uQ.png" /><figcaption>Created by Faith Innocent</figcaption></figure><h3>Introduction</h3><p>My learning journey at DataraFlow continues to deepen my understanding of data analytics and how data can be transformed into meaningful information.</p><p>In Week 8, I focused on grouping and aggregating data, exploring trade datasets using Python and Pandas, writing SQL queries, and learning how to communicate patterns through data visualization.</p><p>This week’s work helped me understand that data analysis is not just about writing code. It is about asking useful questions, organizing information correctly, identifying patterns, and presenting findings in a way that people can understand.</p><h3>Learning to Group and Aggregate Data</h3><p>One of the key topics I explored this week was grouping and aggregating data.</p><p>Real-world datasets often contain many records, making it difficult to understand the bigger picture by looking at individual rows. Grouping allows us to organize records into categories, while aggregation helps us calculate summaries such as totals, counts, averages, and other useful measurements.</p><p>Using Python and Pandas, I practised grouping data by trade flow, commodity type, and partner country.</p><p>These techniques can help answer questions such as:</p><ul><li>What is the total value of imports and exports?</li><li>Which countries are the leading trade partners?</li><li>How do different product categories compare?</li><li>Which partners appear in both import and export records?</li></ul><p>I learned that the quality of the answer depends on how well the data is prepared and how carefully the grouping and aggregation are performed.</p><h3>Exploring International Trade Data</h3><p>Another important part of my assignment involved exploring UK milk and cream trade data.</p><p>The project focused on understanding imports and exports, comparing processed and unprocessed milk products, and examining relationships between the UK and its trading partners.</p><p>To approach this analysis, I worked with fields such as trade flow, commodity code, partner country, period, and trade value.</p><p>An important step was making commodity categories easier to interpret. For example, the dataset uses commodity codes to distinguish between different milk and cream product categories. Translating these codes into readable labels makes the analysis easier to understand.</p><p>I also explored how trade values could be grouped by country and trade flow to identify major partners and compare the different directions of trade.</p><p>This exercise showed me how structured data can help us investigate economic activity and international trade relationships.</p><h3>Understanding the Importance of Data Cleaning</h3><p>Before analyzing any dataset, it is important to understand its structure and check its quality.</p><p>Data can contain unclear labels, unnecessary summary records, inconsistent formats, and other issues that may affect the results.</p><p>During this assignment, I considered the importance of excluding the overall “World” record when calculating totals by individual partner. If an overall total is combined with individual country totals without care, the same trade activity may be counted more than once.</p><p>This reinforced an important lesson: data cleaning is not simply about making a dataset look neat. It is about ensuring that the analysis answers the intended question without introducing avoidable errors.</p><h3>Using SQL to Answer Data Questions</h3><p>Alongside Python and Pandas, Week 8 introduced me to practical SQL analysis.</p><p>SQL provides a structured way to retrieve, group, summarize, and compare data stored in databases.</p><p>The exercises covered queries involving:</p><ul><li>Grouping trade records by flow and product type.</li><li>Comparing monthly imports and exports.</li><li>Calculating trade balances.</li><li>Summarizing quarterly and yearly trade values.</li><li>Ranking leading trade partners.</li><li>Comparing the contributions of different export partners.</li></ul><p>I also explored SQL concepts such as GROUP BY, conditional aggregation, common table expressions, and window functions.</p><p>These techniques are useful because they allow analysts to answer questions directly from a database without always needing to load the entire dataset into Python.</p><p>I am continuing to validate my SQL queries in the DataraFlow course environment to ensure that the results match the intended questions.</p><h3>Exploring Data Visualization</h3><p>Data visualization was another important part of this week’s learning.</p><p>Charts can help us understand patterns that may be difficult to notice in a table of numbers. Different chart types also serve different purposes.</p><p>My assignment focused on five types of visualization:</p><p><strong>1. Histogram</strong></p><p>A histogram helps show how numerical values are distributed. In trade analysis, it can help reveal whether most records have relatively small values while a smaller number have much larger values.</p><p><strong>2. Kernel Density Estimate (KDE) Plot</strong></p><p>A KDE plot provides a smoothed view of a numerical distribution. It can help compare the general shapes of distributions across groups.</p><p><strong>3. Violin Plot</strong></p><p>A violin plot combines information about a distribution’s shape with a comparison between groups. It can help show differences in the spread and concentration of trade values.</p><p><strong>4. Regression Plot</strong></p><p>A regression plot can help explore whether two numerical variables appear to move together. Any relationship observed in a small dataset should be interpreted cautiously rather than treated as proof of causation.</p><p><strong>5. Heatmap</strong></p><p>A heatmap uses colour intensity to make differences across categories easier to see. For this project, a heatmap can help compare monthly import and export totals.</p><p>Working with these visualization methods helped me appreciate the importance of choosing a chart that fits the question being investigated.</p><h3>Challenges and Lessons Learned</h3><p>One of the important lessons from this assignment was that writing code is only one part of data analysis.</p><p>The code must also run correctly, use the appropriate data, and produce results that can be interpreted and checked.</p><p>Working with both Python and SQL also reinforced the need to pay attention to column names, data types, grouping logic, and the order in which calculations are performed.</p><p>I also learned the importance of distinguishing between an analysis that has been prepared and one whose results have been fully verified. Database-dependent queries need to be tested against the actual course database, and visualizations need to be checked using the real dataset before conclusions are drawn.</p><p>These experiences are helping me develop a more careful and responsible approach to data analytics.</p><h3>Key Takeaways from Week 8</h3><p>Here are some of the main lessons I am taking away from this week’s work:</p><ul><li>Grouping and aggregation help turn detailed records into useful summaries.</li><li>Data cleaning is essential for reliable analysis.</li><li>Python and Pandas provide flexible tools for exploring and transforming datasets.</li><li>SQL helps retrieve and summarize information directly from databases.</li><li>Data visualization makes patterns easier to explore and communicate.</li><li>Results should be validated before conclusions are presented as facts.</li><li>Good analysis begins with a clear question and ends with an understandable explanation of the findings.</li></ul><h3>Conclusion</h3><p>Week 8 at DataraFlow has been another valuable step in my journey toward becoming a more confident data analyst.</p><p>Through grouping and aggregation, trade data exploration, SQL practice, and data visualization, I am learning how to approach datasets more systematically and connect technical skills to practical questions.</p><p>I understand that becoming proficient in data analytics requires consistent practice, patience, and a willingness to investigate mistakes rather than ignore them.</p><p>As I continue learning, my goal is to strengthen my technical skills, improve the accuracy of my analysis, and communicate data-driven insights more effectively.</p><p>Every week is an opportunity to learn something new, build on previous knowledge, and move closer to my goals.</p><p>I look forward to continuing this journey and applying what I learn to real-world problems.</p><p><strong>Thank you for reading about my Week 8 learning experience at DataraFlow.</strong></p><p>If you are also learning Python, SQL, or data analytics, I would love to connect, learn from your experiences, and exchange ideas.</p><p>#DataraFlow #DataAnalysis #Python #Pandas #SQL #DataVisualization #DataCleaning #LearningInPublic</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5ce1f3e8fa00" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Week 7 at DataraFlow: Cleaning Data, Finding Patterns, and Combining Datasets]]></title>
            <link>https://medium.com/@faithinnocent774/week-7-at-dataraflow-cleaning-data-finding-patterns-and-combining-datasets-4e482e680187?source=rss-5fdc0e3bfa43------2</link>
            <guid isPermaLink="false">https://medium.com/p/4e482e680187</guid>
            <category><![CDATA[python]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data-analyst]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[learning]]></category>
            <dc:creator><![CDATA[Faith  Innocent]]></dc:creator>
            <pubDate>Sat, 24 Oct 2026 19:40:42 GMT</pubDate>
            <atom:updated>2026-10-24T19:40:42.829Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Week 7 at DataraFlow: Cleaning Data, Finding Patterns, and Combining Datasets</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f5s1u2gEjt6DOC1CJCmGZg.png" /><figcaption>Created by Faith Innocent</figcaption></figure><p><strong>Introduction</strong></p><p>My Week 7 learning journey at DataraFlow focused on an important part of data analysis: preparing data before using it to answer questions.</p><p>I explored data cleaning, data visualisation, data transformation, and combining datasets using Python, Pandas, and SQL. Through the weekly exercises and projects, I learnt that data analysis involves more than writing code. It also requires understanding the data, identifying problems, selecting appropriate techniques, and interpreting results carefully.</p><p>In this article, I share the concepts I studied, the projects I worked on, the challenges I encountered, and the lessons I am taking forward.</p><p><strong>1. What I Learnt This Week</strong></p><p>Data Cleaning with Pandas</p><p>One of my key lessons was understanding why data cleaning is necessary before analysis.</p><p>Real-world datasets may contain missing values, duplicate records, inconsistent column names, dates stored as text, or numeric values mixed with unwanted characters. These issues can affect calculations and produce misleading results if they are not addressed.</p><p>I practised using several Pandas functions:</p><p>- &quot;read_csv()&quot; to load data from CSV files.<br>- &quot;head()&quot; and &quot;shape&quot; to inspect datasets.<br>- &quot;to_datetime()&quot; to convert date values into a datetime format.<br>- &quot;to_numeric()&quot; to convert values into numbers.<br>- &quot;dropna()&quot; to remove records with missing values where appropriate.<br>- &quot;drop_duplicates()&quot; to identify and remove duplicate records.<br>- Boolean filtering to select records that meet particular conditions.</p><p>I also learnt the importance of inspecting column names and sample records before deciding how to clean a dataset.</p><p>Data Visualisation</p><p>I explored how visualisation can make patterns in data easier to understand.</p><p>Using Pandas plotting and Matplotlib, I worked with techniques for displaying temperature trends and examining relationships between numerical variables.</p><p>Visualisation is useful because a table of numbers may not immediately reveal a pattern. A graph can help identify trends, unusual observations, and differences between groups.</p><p>However, I also learnt that a visual pattern should be interpreted carefully rather than treated as proof of a particular explanation.</p><p>Data Transformation and Combining Datasets</p><p>Another important part of the week was transforming data into a more useful format and combining related datasets.</p><p>I worked with World Bank datasets containing information about countries&#39; gross domestic product (GDP) and life expectancy.</p><p>The process involved inspecting the data, selecting relevant columns, handling missing values, transforming GDP into a more readable unit, rounding life expectancy values, and merging the datasets using a shared country column.</p><p>I also explored the Spearman rank correlation coefficient as a way to investigate whether two variables tend to move together.</p><p>One important lesson is that correlation measures association, not causation. Even when two variables are correlated, that does not establish that one causes the other.</p><p><strong>2. Project A: Holiday Weather Analysis</strong></p><p>The first major project focused on historical weather data for London Heathrow.</p><p>The goal was to use weather observations to identify potentially suitable seven-day periods for a summer holiday.</p><p>The project involved obtaining historical weather data, loading it into a Pandas DataFrame, cleaning the data, and selecting the summer months of June, July, and August.</p><p>I prepared a workflow to convert date values into a datetime format, handle unwanted characters in numerical fields, inspect missing values, and filter the data by month.</p><p>The next step was to compare seven-day periods using rainfall and maximum temperature. A practical approach is to consider periods with lower total rainfall and higher average maximum temperatures.</p><p>This project helped me understand how data analysis can support everyday decisions. Rather than relying entirely on assumptions about the weather, historical observations can provide useful information for comparing possible holiday dates.</p><p>Historical data cannot guarantee future weather conditions, but it can help us explore past patterns and make more informed plans.</p><p>An important next step: The actual London Heathrow 2023 CSV data must be loaded and analysed before specific holiday dates can be recommended. I will only report the final dates after checking the real results.</p><p><strong>3. Project B: GDP and Life Expectancy</strong></p><p>The second major project explored the relationship between a country&#39;s GDP and the life expectancy of its population.</p><p>For this analysis, the project uses two World Bank datasets: one containing GDP figures and another containing life expectancy figures.</p><p>I worked through the main stages of preparing the datasets for comparison:</p><p>1. Inspecting the datasets to understand their structure.<br>2. Removing aggregate entries and handling missing values.<br>3. Converting GDP from US dollars into millions of British pounds using the conversion rate specified in the exercise.<br>4. Rounding life expectancy values for easier interpretation.<br>5. Selecting the columns required for the analysis.<br>6. Combining the cleaned datasets using an inner join on the country column.</p><p>7.Exploring the relationship between GDP and life expectancy using Spearman correlation and a scatterplot.</p><p>I also examined the idea of comparing countries with the lowest GDP against countries with the lowest life expectancy.</p><p>This comparison highlights an important limitation of using total GDP as a measure of how well people live. Countries have different population sizes, so a country&#39;s total economic output does not tell us how much income is available per person.</p><p>GDP per capita may provide a more useful comparison of average economic output per person, although it still does not show how income is distributed within a country.</p><p>The project reinforced the importance of choosing suitable indicators and considering the limitations of the data before drawing conclusions.</p><p><strong>4. My Own World Bank Indicators Project</strong></p><p>The assignment also challenged me to develop a separate project using two World Bank indicators of my choice.</p><p>For my proposed project, I selected:</p><p><strong>- Internet usage:</strong> Individuals using the Internet, as a percentage of the population (&quot;IT.NET.USER.ZS&quot;).<br><strong>- GDP per capita:</strong> GDP per capita in current US dollars (&quot;NY.GDP.PCAP.CD&quot;).</p><p>I selected these indicators because I wanted to explore whether countries with higher GDP per capita also tend to have higher levels of Internet usage.</p><p>The proposed analysis uses data for 2022. It involves retrieving the indicator data, matching countries, handling missing or invalid values, calculating Spearman correlation, and creating a scatterplot.</p><p>This is an interesting question because Internet access can be influenced by several factors, including affordability, infrastructure, education, and government policy. Economic output per person is only one possible part of the explanation.</p><p>The analysis would help me investigate whether there is an association between the two indicators across countries.</p><p>However, the World Bank API request did not complete successfully in my initial run, so I still need to resolve the data-retrieval issue and calculate the actual correlation before reporting a result.</p><p><strong>5. Learning SQL for Data Cleaning and Analysis</strong></p><p>Another important part of Week 7 was practising SQL.</p><p>While Pandas allows us to clean and analyse data in Python, SQL enables us to query and transform data directly in a database.</p><p>I worked on five SQL practice tasks.</p><p><strong>Task 1: Cleaning a Messy Column</strong></p><p>The first task involved a wind-direction column containing unwanted HTML line-break text.</p><p>The solution uses SQL functions such as &quot;REPLACE()&quot; and &quot;TRY_CAST()&quot; to remove the unwanted characters, convert the remaining value to an integer, and count observations where the wind direction exceeds 350 degrees.</p><p>This exercise showed me how data cleaning can be performed directly in SQL.</p><p><strong>Task 2: Filtering a Date Range</strong></p><p>The second task focused on retrieving weather observations between 8 December and 12 December 2014.</p><p>The query converts the text-based date column into a date value, filters the requested period, and sorts the results chronologically.</p><p>This helped reinforce the importance of handling dates correctly when filtering time-based data.</p><p><strong>Task 3: Combining Tables with JOIN</strong></p><p>The third task involved joining the GDP and life expectancy tables using the country column.</p><p>The query filters countries with GDP below 10 billion US dollars and a known life expectancy, then selects the ten entries with the highest life expectancy.</p><p>I also practised rounding GDP into billions of US dollars and life expectancy to one decimal place.</p><p><strong>Task 4: Understanding What Is Lost in a Join</strong></p><p>The fourth task examined how missing values and unmatched records affect a combined dataset.</p><p>Using a &quot;LEFT JOIN&quot; and conditions that check for missing values, the query counts GDP records that do not have usable matching information.</p><p>This demonstrated why it is important to understand what happens to records when datasets are cleaned and combined.</p><p><strong>Task 5: Finding the Driest Summer Week</strong></p><p>The final SQL task involved finding the seven-day period with the lowest total rainfall during June, July, and August 2014.</p><p>The intended approach uses a common table expression and a SQL window function to calculate rainfall totals over seven consecutive rows, then selects the driest eligible period.</p><p>This introduced me to window functions and showed how SQL can support time-based analysis.</p><p>These exercises helped me understand how SQL and Python can complement one another in a data analysis workflow.</p><p>Verification note: I prepared the SQL queries and ran them in a mock practice environment. That environment confirmed that the cells executed, but it did not validate the queries against DataraFlow&#39;s real database or verify the expected answers. The final results still need to be checked in the actual course environment.</p><p><strong>6. Challenges I Encountered</strong></p><p>One challenge was working with inconsistent data formats. A value that looks numerical may contain additional characters, while a date may initially be stored as ordinary text.</p><p>Another challenge was understanding how cleaning decisions affect the final analysis. Removing missing values can make calculations easier, but it may also reduce the number of records available for comparison.</p><p>I also encountered a problem retrieving data from the World Bank API. Without successfully loading the indicator data, I could not calculate and interpret the final correlation for my own project.</p><p>Finally, I learnt that a successful code execution does not necessarily mean that an analysis is correct. The results must be checked against the real data and the requirements of the task.</p><p><strong>7. How I Plan to Overcome These Challenges</strong></p><p>To improve my approach, I plan to follow a more systematic process:</p><p>- Inspect datasets before making changes.<br>- Check column names, data types, missing values, and duplicate records.<br>- Clean and transform data in clearly defined steps.<br>- Check the number of rows before and after cleaning or merging.<br>- Test SQL queries against the actual database rather than relying only on mock results.<br>- Confirm that external data sources have loaded successfully before interpreting results.<br>- Use tables, visualisations, and statistical measures together when exploring relationships.<br>- Document assumptions and limitations so that readers can understand how conclusions were reached.</p><p>These steps will help me produce analyses that are more reliable, transparent, and easier to reproduce.</p><p><strong>8. My Key Takeaways</strong></p><p>Week 7 helped me appreciate that good data analysis begins long before the final chart or conclusion.</p><p>Cleaning data carefully, selecting appropriate variables, transforming values correctly, and combining datasets using suitable keys are all essential steps.</p><p>I also learnt that SQL and Pandas provide different but complementary ways to work with data. SQL can retrieve and transform information in a database, while Python provides flexible tools for cleaning, statistical analysis, and visualisation.</p><p>Most importantly, I learnt to be cautious when interpreting results. A correlation does not establish causation, and the quality of a conclusion depends on the quality of the data and the methods used.</p><p><strong>Conclusion</strong></p><p>My Week 7 experience at DataraFlow strengthened my understanding of data cleaning, visualisation, transformation, dataset merging, and SQL.</p><p>The weather project showed how historical observations can support practical decisions, while the GDP and life expectancy project demonstrated the importance of choosing appropriate indicators and interpreting relationships carefully.</p><p>I still have steps to complete before I can report verified results for every part of the assignment, particularly loading the actual weather data, resolving the World Bank API issue, and testing the SQL queries against the real database.</p><p>As I continue learning, I want to become more confident in moving from raw data to meaningful analysis and communicating my findings clearly.</p><p>Every challenge is an opportunity to learn something new, and I look forward to applying these skills to more practical data projects.</p><p>#DataraFlow #DataAnalysis #Python #Pandas #SQL #DataCleaning #DataVisualization #WorldBank #LearningInPublic</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4e482e680187" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Predicting Administrative Task Delays Using Machine Learning]]></title>
            <link>https://medium.com/@faithinnocent774/predicting-administrative-task-delays-using-machine-learning-f87881784a5a?source=rss-5fdc0e3bfa43------2</link>
            <guid isPermaLink="false">https://medium.com/p/f87881784a5a</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[predictive-analytics]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Faith  Innocent]]></dc:creator>
            <pubDate>Thu, 08 Oct 2026 11:42:19 GMT</pubDate>
            <atom:updated>2026-10-08T11:42:19.687Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Abstract</strong></p><p>Administrative task delays can affect productivity, service delivery, workload planning, and organizational efficiency. This study investigates whether machine learning can be used to predict administrative task delays using historical help desk workflow records.</p><p>The study uses a dataset containing <strong>66,691 task records and 58 variables</strong>. Because the dataset does not contain an official deadline or service-level agreement (SLA), a study-defined rule was used: within each calendar year, tasks whose workflow processing time was above the 75th percentile were classified as delayed, while the remaining tasks were classified as on time.</p><p>Three machine learning models were developed and compared: Logistic Regression, Decision Tree, and Random Forest. Categorical variables such as task type, priority, project, month, day of the week, and creation hour were used as predictors. The models were evaluated using accuracy, precision, recall, F1-score, and ROC-AUC.</p><p>Random Forest produced the strongest overall performance, achieving <strong>74.95% accuracy and 0.807 ROC-AUC</strong> on the holdout test set. Five-fold cross-validation produced an average accuracy of approximately <strong>75.13%</strong> and ROC-AUC of <strong>0.803</strong>.</p><p>The findings suggest that machine learning can provide useful early warning signals for tasks that may experience delays. Such predictions could support workload planning, task allocation, and administrative decision-making. However, the model should be treated as a decision-support tool rather than an automatic decision-maker.</p><p><strong>Keywords:</strong> Administrative Task Delays, Machine Learning, Random Forest, Logistic Regression, Decision Tree, Workflow Management, Predictive Analytics</p><ol><li><strong>Introduction</strong></li></ol><p>Administrative work often involves multiple tasks that must be completed within reasonable timeframes. Delays in processing tasks can increase workloads, affect service delivery, and reduce operational efficiency. Being able to identify tasks that are likely to experience delays before the delay occurs could help organizations plan resources and take preventive action.</p><p>Machine learning provides a way to learn patterns from historical data and use those patterns to make predictions about new cases. In an administrative environment, historical task records can contain information about task type, priority, project context, and the time at which a task was created. These characteristics may provide useful signals about whether a task is likely to experience a longer processing period.</p><p>This study therefore examines whether machine learning can predict administrative task delays using historical workflow records. Three commonly used classification approaches — Logistic Regression, Decision Tree, and Random Forest — are compared to determine which provides the most useful predictive performance.</p><p><strong>2. Problem Statement</strong></p><p>Administrative task delays can be difficult to anticipate because they may depend on several factors, including the type of task, its priority, the project involved, and when the task enters the workflow.</p><p>Without a predictive approach, organizations may only recognize a delay after it has already occurred. This can make it harder to redistribute workloads, prioritize tasks, and respond to potential bottlenecks.</p><p>This study addresses this problem by developing machine learning models that classify administrative workflow records as either On Time or Delayed. The study compares different machine learning algorithms to determine whether historical workflow characteristics can provide useful predictions of task delays.</p><p><strong>3. Aim and Objectives</strong></p><p>The main aim of this study is to determine whether machine learning can predict administrative task delays.</p><p>The specific objectives are to:</p><p>1. Prepare and clean historical administrative workflow data for analysis.<br>2. Define an operational classification of tasks as On Time or Delayed.<br>3. Identify workflow characteristics associated with task delays.<br>4. Develop Logistic Regression, Decision Tree, and Random Forest classification models.<br>5. Compare the models using accuracy, precision, recall, F1-score, and ROC-AUC.<br>6. Identify the model that provides the strongest overall predictive performance.<br>7. Discuss how delay predictions could support administrative planning and decision-making.</p><p><strong>4. Research Questions</strong></p><p>This study addresses the following questions:</p><p>1. Can machine learning predict whether an administrative task will be delayed?<br>2. Which workflow characteristics are associated with administrative task delays?<br>3. Which of the three tested machine learning models provides the strongest predictive performance?<br>4. How could delay predictions support workload planning and administrative decision-making?</p><p><strong>5. Dataset and Variables</strong></p><p>The study uses a Help Desk Tickets dataset containing 66,691 records and 58 variables. The dataset contains historical workflow information about tasks, including task type, priority, project context, creation date, resolution date, and other workflow-related characteristics.</p><p>The dataset was examined and cleaned using Python. Date fields were converted into appropriate datetime formats, and the relevant variables were checked for missing values and consistency.</p><p>The dataset does not contain an official organizational deadline or service-level agreement (SLA) that could directly determine whether a task was late. Therefore, this study uses a study-defined operational rule based on workflow processing time.</p><p>For each calendar year, the 75th percentile of workflow processing time was calculated. A task with processing time above that yearly threshold was classified as Delayed, while a task at or below the threshold was classified as On Time.</p><p>Using this approach, 16,675 records (25.00%) were classified as Delayed and 50,016 records (75.00%) were classified as On Time.</p><p>The predictive variables used in the final models were:</p><p>- Issue type<br>- Issue priority<br>- Project/context<br>- Month of task creation<br>- Day of the week of task creation<br>- Hour of task creation</p><p>Variables such as total workflow processing time were not used as predictors because processing time was used to define the outcome. Other workflow variables that can accumulate during task processing were also excluded to reduce the risk of information leakage.</p><p>The dataset does not contain direct measurements for staff experience or number of documents processed. These variables were therefore not invented or estimated and were not included in the final model.</p><p><strong>6. Methodology</strong></p><p>The analysis was conducted using Python and commonly used machine learning techniques.</p><p><strong>6.1 Data Preparation</strong></p><p>The dataset was loaded into Python and inspected for its structure, data types, missing values, and consistency. Date variables were converted to datetime format, allowing the extraction of month, day of the week, and hour of task creation.</p><p>Categorical variables were converted into numerical representations using one-hot encoding so that they could be used by the machine learning algorithms.</p><p><strong>6.2 Outcome Definition</strong></p><p>Because there was no explicit deadline field, the study defined delay using the yearly 75th percentile of workflow processing time.</p><p>Tasks above the relevant yearly threshold were labelled Delayed, while tasks at or below the threshold were labelled On Time.</p><p>This definition is an analytical rule for this study and should not be interpreted as an official organizational deadline.</p><p><strong>6.3 Machine Learning Models</strong></p><p>Three classification models were developed:</p><p>Logistic Regression was used as a simple and interpretable baseline model.</p><p>Decision Tree was used to capture nonlinear relationships between task characteristics and delay outcomes.</p><p>Random Forest was used as an ensemble model capable of combining multiple decision trees to improve predictive performance.</p><p>The data was divided into training and testing sets using an 80:20 stratified split. Stratification was used to preserve the proportion of delayed and on-time tasks in both sets.</p><p><strong>6.4 Model Evaluation</strong></p><p>The models were evaluated using five main performance measures:</p><p>Accuracy measures the proportion of all predictions that were correct.</p><p>Precision measures how many tasks predicted as delayed were actually delayed.</p><p>Recall measures how many of the truly delayed tasks were correctly identified.</p><p>F1-score provides a balance between precision and recall.</p><p>ROC-AUC measures how well the model distinguishes between delayed and on-time tasks across different classification thresholds.</p><p>Five-fold stratified cross-validation was also used to assess whether the model performance was reasonably consistent across different subsets of the data.</p><p><strong>7. Results</strong></p><p>The analysis compared the performance of Logistic Regression, Decision Tree, and Random Forest models.</p><p><strong>7.1 Model Performance</strong></p><p>The results from the holdout test set were as follows:</p><p>Model | Accuracy | Precision | Recall | F1-score | ROC-AUC</p><p>Logistic Regression | 69.2% | 43.0% | 71.8% | 53.8% | 0.757</p><p>Decision Tree | 73.1% | 47.3% | 66.6% | 55.3% | 0.794</p><p>Random Forest | 75.0% | 49.9% | 63.5% | 55.9% | 0.807</p><p>Random Forest produced the strongest overall performance. It achieved approximately 75.0% accuracy and the highest ROC-AUC of 0.807. It also produced the highest F1-score among the three models.</p><p>The five-fold cross-validation results were consistent with the holdout test results. Random Forest achieved an average accuracy of approximately 75.1% and an average ROC-AUC of approximately 0.803, indicating that its performance was reasonably stable across different subsets of the data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qHXAND4TAhVw5JeskVD0HQ.png" /><figcaption>Created by Faith Innocent</figcaption></figure><p><strong>7.2 Factors Associated With Delays</strong></p><p>The descriptive analysis showed differences in delay rates across task characteristics.</p><p>Task type showed a noticeable relationship with delays. Epic and Project records had substantially higher delay rates than several other task categories, while Assistance and Vacation records had relatively low delay rates.</p><p>Priority also showed differences in delay rates. Lower-priority tasks generally had higher observed delay rates than higher-priority tasks in this dataset.</p><p>The analysis also found that the month in which a task was created showed a stronger association with the delay classification than several of the other individual categorical variables. This indicates that workflow timing and operational patterns may be important when predicting delays.</p><p>These associations do not mean that a particular task type or priority directly causes a delay. They indicate patterns present in the historical dataset that can be useful for prediction.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Rssid_rbPLJfMnKc8rRleA.png" /><figcaption>Created by Faith Innocent</figcaption></figure><p><strong>7.3 Random Forest Confusion Matrix</strong></p><p>On the holdout test set, the Random Forest model correctly identified 2,118 delayed tasks and 7,880 on-time tasks.</p><p>It incorrectly classified 1,217 delayed tasks as on time and incorrectly classified 2,124 on-time tasks as delayed.</p><p>The recall of approximately 63.5% means that the model identified about two-thirds of the tasks classified as delayed. Its precision of approximately 49.9% means that about half of the tasks predicted as delayed were actually in the delayed category.</p><p><strong>8. Discussion</strong></p><p>The results indicate that machine learning can provide useful predictive information about administrative task delays.</p><p>Among the three models tested, Random Forest performed best overall. Its ROC-AUC of 0.807 indicates good discrimination between the two study-defined outcome classes. Its performance was also relatively stable during five-fold cross-validation.</p><p>The Logistic Regression model provided a useful baseline but performed less strongly than the tree-based models. The Decision Tree improved on Logistic Regression, while Random Forest achieved the strongest overall results.</p><p>The results also show that prediction of delays is not perfect. The Random Forest model missed some delayed tasks and generated some false delay predictions. Therefore, the model should not be used as a replacement for human judgment.</p><p>The findings suggest that administrative organizations could use predictive models as an early-warning mechanism. For example, tasks identified as having a higher probability of delay could receive additional attention, earlier follow-up, or workload adjustments.</p><p>However, the results should be interpreted carefully because the delay outcome was defined using a study-specific percentile rule rather than an official organizational deadline.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eiToyoSlGum1eT2ODCCHhg.png" /><figcaption>Created by Faith Innocent</figcaption></figure><p><strong>9. Practical Implications</strong></p><p>A predictive system based on this approach could support administrative teams in several ways.</p><p>First, it could help managers identify tasks that may require additional attention before they become severely delayed.</p><p>Second, predictions could support workload planning by highlighting periods or categories associated with higher delay risk.</p><p>Third, the system could assist with task prioritization and resource allocation.</p><p>Finally, regular monitoring of model predictions could help organizations identify changing workflow patterns and improve administrative processes over time.</p><p>The model should be used as a decision-support tool. Final decisions should remain with responsible administrative staff who can consider information that may not be available in the dataset.</p><p><strong>10. Limitations</strong></p><p>This study has several limitations.</p><p>First, the dataset does not contain an official deadline or service-level agreement. The delayed/on-time classification therefore relies on a study-defined yearly 75th-percentile threshold.</p><p>Second, the dataset does not contain direct variables for staff experience or number of documents processed, although these factors may influence administrative delays.</p><p>Third, some workflow variables were excluded from prediction because they could contain information that becomes available only after a task has progressed through the workflow. Excluding these variables reduces information leakage but also limits the available predictive information.</p><p>Fourth, the dataset represents a particular help desk workflow environment. The findings may therefore not generalize directly to every administrative organization.</p><p>Finally, future studies should evaluate the models using time-ordered validation on newer records to determine how well the approach performs when predicting genuinely future tasks.</p><p><strong>11. Conclusion</strong></p><p>This study investigated whether machine learning can predict administrative task delays using historical workflow data.</p><p>Three classification models—Logistic Regression, Decision Tree, and Random Forest—were developed and evaluated. Random Forest achieved the strongest overall performance, with approximately 75.0% holdout accuracy and a ROC-AUC of 0.807.</p><p>The findings demonstrate that historical task characteristics can provide useful signals for identifying tasks that are more likely to fall into a delayed processing category. However, prediction is not perfect, and the results should be used to support rather than replace administrative judgment.</p><p>The study provides a practical starting point for applying machine learning to administrative workflow management. Future research could improve the approach by incorporating explicit deadlines, staff workload, staff experience, document counts, and other operational variables, as well as testing the model on future time periods.</p><p><strong>12. References</strong></p><p>Breiman, L. (2001). Random forests. Machine Learning, 45, 5–32. https://doi.org/10.1023/A:1010933404324</p><p>scikit-learn. LogisticRegression documentation. https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html</p><p>scikit-learn. Model evaluation documentation. https://scikit-learn.org/stable/modules/model_evaluation.html</p><p>scikit-learn. Getting Started documentation. https://scikit-learn.org/stable/getting_started</p><p>Help Desk Tickets Dataset. Dataset documentation and feature descriptions.</p><p><strong>13. Reproducibility Summary</strong></p><p>The analysis was conducted using Python with pandas and scikit-learn. The workflow included data loading, data validation, datetime processing, creation of the study-defined delay outcome, categorical encoding, stratified train-test splitting, model training, cross-validation, and evaluation using classification metrics.</p><p>A fixed random state of 42 was used for the train-test split and machine learning models where applicable. This supports reproducibility of the reported analysis.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f87881784a5a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Week 4: From an Idea to an Approved Research Topic]]></title>
            <link>https://medium.com/@faithinnocent774/week-4-from-an-idea-to-an-approved-research-topic-e6190dcf523d?source=rss-5fdc0e3bfa43------2</link>
            <guid isPermaLink="false">https://medium.com/p/e6190dcf523d</guid>
            <category><![CDATA[research]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Faith  Innocent]]></dc:creator>
            <pubDate>Mon, 05 Oct 2026 11:19:46 GMT</pubDate>
            <atom:updated>2026-10-05T11:19:46.101Z</atom:updated>
            <content:encoded><![CDATA[<p>Week 4 was an important stage in my DataraFlow learning journey. Unlike the previous weeks, which focused mainly on developing my technical skills in Python and data science, this week challenged me to think about a real-world problem that I could investigate through research.</p><p>As I continue learning and developing my skills, I am becoming more interested in how technology can solve everyday problems, improve productivity, and make work more efficient. This encouraged me to think about how I could connect my interests, experience, and growing knowledge of data science to a practical research topic.</p><h3>Choosing My Research Topic</h3><p>For my research, I proposed the topic <strong>“The Role of Artificial Intelligence in Improving Administrative Efficiency.”</strong></p><p>I chose this topic because administrative work is an important part of how organizations operate. Tasks such as scheduling, data entry, document management, customer communication, and information processing require time, organization, and attention to detail.</p><p>From my interest in administrative and virtual assistance work, I understand the importance of completing these tasks accurately and efficiently. I also recognize that technology is changing how people work, and artificial intelligence is becoming an important tool for improving productivity.</p><p>This made me interested in exploring how AI tools can support administrative professionals, reduce repetitive work, and help organizations manage their daily operations more effectively.</p><figure><img alt="By FAKURIANDESIGN" src="https://cdn-images-1.medium.com/max/1024/0*gynaTba3QmrkuiPh" /><figcaption>Photo by Faith Innocent</figcaption></figure><h3>Understanding the Problem Behind the Topic</h3><p>One thing I realized during this process is that choosing a research topic involves more than finding a title that sounds interesting. A good research topic should identify a problem worth investigating and explain why it matters.</p><p>For this study, I want to examine how artificial intelligence can improve administrative efficiency in organizations. I am particularly interested in how AI can assist with scheduling, data entry, document management, customer communication, and information processing.</p><p>I also want to consider the potential benefits of using AI, including time savings, improved accuracy, faster decision-making, and increased productivity.</p><p>However, I understand that AI adoption also comes with challenges. Issues such as data privacy, implementation costs, and the need for employee training must be considered. Examining both the benefits and the challenges will help me develop a more balanced understanding of the topic.</p><h3>Connecting My Interests With Data Science</h3><p>This research idea has helped me see how my learning in data science can connect with practical workplace challenges.</p><p>Data science is not only about writing code or learning technical tools. It also involves asking meaningful questions, examining available information, identifying patterns, and using evidence to develop a better understanding of a problem.</p><p>For this research, I plan to use existing literature and relevant data to investigate how AI can support administrative professionals and improve organizational operations.</p><p>This approach will help me explore the relationship between technology and workplace productivity while continuing to build my research and analytical skills.</p><figure><img alt="plexus net" src="https://cdn-images-1.medium.com/max/1024/0*nqyc5mmx4LTDXsZt" /><figcaption>Photo by Faith Innocent</figcaption></figure><h3>Getting My Research Topic Approved</h3><p>One of the most encouraging moments of Week 4 was receiving approval for my proposed research topic.</p><p>Seeing my topic marked as approved on the DataraFlow platform was an important milestone. It gave me a clear direction for my research and allowed me to move from developing an idea to preparing for the next stage of the process.</p><p>Although the approval was a positive step, I understand that it is only the beginning. I still need to develop the research further, study relevant literature, identify suitable data sources, and organize my findings around the research objectives.</p><p>I look forward to working through these stages carefully and improving my ability to investigate a problem using evidence.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/929/1*s7gOwYighkpVbgzULykBsA.png" /><figcaption>Created by Faith Innocent</figcaption></figure><h3>What Week 4 Taught Me</h3><p>The biggest lesson I learned during Week 4 is that learning data science involves more than mastering programming languages and technical concepts. It also requires critical thinking, problem identification, research, and the ability to connect technical knowledge to real-world situations.</p><p>Choosing this topic encouraged me to think about how artificial intelligence could influence administrative work and how organizations might use technology responsibly to improve efficiency.</p><p>It also reminded me that research requires a balanced perspective. While AI offers opportunities to save time and improve productivity, its challenges must also be understood before making conclusions about its effectiveness.</p><p>Receiving approval for my research topic has motivated me to continue learning and take the next steps with greater focus.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/941/1*u_MnQry3eEZyL36LUhE-7Q.png" /><figcaption>Crested by Faith Innocent</figcaption></figure><h3>Looking Ahead</h3><p>My next step is to begin developing the research beyond the approved title. I intend to explore existing literature, identify relevant information and data sources, refine my research questions, and build a clearer understanding of how AI can improve administrative efficiency.</p><p>I know there is still much to learn, but I am excited to continue developing my Python, data analysis, and research skills through this process.</p><p>Week 4 has reminded me that progress often begins with asking the right questions and being willing to investigate them.</p><p><strong>One research topic approved, one step closer to turning an idea into meaningful research.</strong></p><p>I look forward to continuing this journey and discovering what I can learn along the way.</p><p>#ArtificialIntelligence #AI #DataScience #DataAnalytics #AdministrativeEfficiency #DigitalTransformation #Productivity #ResearchJourney #Python #DataraFlow #LearningJourney #FutureOfWork</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e6190dcf523d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Public Administration to Python: How I Went From Confused to Confident in One Week]]></title>
            <link>https://medium.com/@faithinnocent774/from-public-administration-to-python-how-i-went-from-confused-to-confident-in-one-week-d9bf43565fb2?source=rss-5fdc0e3bfa43------2</link>
            <guid isPermaLink="false">https://medium.com/p/d9bf43565fb2</guid>
            <dc:creator><![CDATA[Faith  Innocent]]></dc:creator>
            <pubDate>Sun, 27 Sep 2026 11:28:09 GMT</pubDate>
            <atom:updated>2026-10-05T12:13:50.259Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Week 3 From Public Administration to Python:</strong> How I Went From Confused to Confident in One Week</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*VXp5H-CjwKAQqg35" /><figcaption>Photo by Faith Innocent</figcaption></figure><p><strong>My Week 3 Journey Learning Python Modules, Error Handling, and Google Colab</strong></p><p>I never thought a Public Administration student from ABU DLC Zaria would be writing Python code.</p><p>A month ago, if you asked me what pip or venv meant, I would have guessed they were some new social media apps. Today, I just completed 30 Python practice exercises in Google Colab and published my first notebook.</p><p>This is my learning journey.</p><p><strong>What I Learned</strong></p><p>Week 3 was all about Python Modules — the building blocks that make Python powerful.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*iErWsc6JUivEUpoK" /><figcaption>Photo by Faith Innocent</figcaption></figure><p><strong>I learned:</strong><br>- math module: How to calculate square roots http://math.sqrt(144), factorials, GCD, and use http://math.pi to calculate the area of a circle.<br>- datetime module: How to get the current time, format dates as DD-MM-YYYY, and build a Date Calculator that tells you how many days are left to your next birthday.<br><strong>- json module:</strong> How to turn my Python dictionary {&quot;name&quot;: &quot;Faith&quot;, &quot;age&quot;: 21, &quot;grade&quot;: &quot;A&quot;} into a JSON string and back — something I now know is used everywhere in real apps.<br><strong>- Error handling and user input:</strong> How to stop my program from crashing using try/except when someone divides by zero or types &quot;abc&quot; instead of a number.<br><strong>- pip and virtual environments:</strong> This was the most technical part. I learned how to create an isolated environment with python -m venv myenv and install packages like requests and numpy without breaking my main system.</p><p>My two favorite assignments were the Math Quiz Game that asks 5 random questions and keeps score, and the JSON File Handling where I saved a list of student records into a file.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*iDwCtNvBb_LMLZJ2" /><figcaption>Photo by Faith Innocent</figcaption></figure><p><strong>The Challenges I Faced</strong></p><p>It wasn&#39;t smooth.</p><p>My biggest challenge was virtual environments in Google Colab. The normal command failed. I had to delete the environment, create it with --without-pip, manually download http://get-pip.py, and then install requests using /content/myenv/bin/pip install requests. That took me almost 2 hours to figure out.</p><p>The second challenge was error handling. My code kept raising JSONDecodeError and ValueError. I learned to read the red error text instead of panicking.</p><p>How It Changed My Perspective</p><p>This week changed everything for me.</p><p>As a Public Administration student, I used to see data as just tables in a textbook. Now I see it as something I can control. I can write a program that calculates age from birth year, checks if a number is positive or negative, and calculates days between two dates — all useful for admin work.</p><p>Google Colab made it easy. You don&#39;t need a powerful laptop. You just need Gmail and internet. You open <strong>http://colab.research.google.com</strong> and you are ready to code.</p><p>If you are a non-tech student like me, my advice is simple: start. Run your first print(&quot;Hello World&quot;) today. You will be shocked how far you can go in one week.</p><p>You can view my full practice notebook here: <strong>https://colab.research.google.com/drive/1MvwIlFlnyP_5QtPfHMN1EIiScjPJGHNr?usp=sharing</strong></p><figure><img alt="My Calculator" src="https://cdn-images-1.medium.com/max/1024/0*DLuP8x33XPHrEYJc" /><figcaption>Photo by Faith Innocent</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d9bf43565fb2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Week 2: Building Faith Community Library with Python OOP]]></title>
            <link>https://medium.com/@faithinnocent774/my-week-2-building-faith-community-library-with-python-oop-ea93787966b9?source=rss-5fdc0e3bfa43------2</link>
            <guid isPermaLink="false">https://medium.com/p/ea93787966b9</guid>
            <dc:creator><![CDATA[Faith  Innocent]]></dc:creator>
            <pubDate>Wed, 16 Sep 2026 15:33:42 GMT</pubDate>
            <atom:updated>2026-09-16T15:33:42.573Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Code On" src="https://cdn-images-1.medium.com/max/1024/0*kUiHG8nZlNrxPuFn" /><figcaption>Photo by Faith Innocent</figcaption></figure><p><strong>Introduction</strong><br>This week in DataraFlow, we moved deeper into Python with Object-Oriented Programming. I built a Library Management System to practice everything I learned.</p><p><strong>What I Learnt</strong></p><p><strong>1. Inheritance</strong><br>I learnt that a child class can reuse properties and methods from a parent class. In my project, EBook and PrintedBook inherit from Book, and StudentMember and TeacherMember inherit from Member using super(). This helped me avoid rewriting code for title, author, year, name and member_id.</p><p><strong>2. Polymorphism</strong><br>Polymorphism means the same method name can behave differently depending on the object. I implemented get_description() differently for EBook (shows file_size and format) and PrintedBook (shows weight and dimensions), and borrow_book() with different limits — 3 books for StudentMember and 5 books for TeacherMember.</p><p><strong>3. Iterators</strong><br>I learnt about __iter__() and __next__() and how StopIteration tells Python when there are no more values to return. I made Library iterable so I can loop through all books with `for book in library` and Member iterable to loop through borrowed books.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*IJBs7-oFaYYedKgr" /><figcaption>Photo by Faith Innocent</figcaption></figure><p><strong>4. Scope &amp; Encapsulation</strong><br>I learnt about local and global variables and how scope works. In my project, I used protected _borrowed_books in Member and private __library_name in Library with get_library_name() and set_library_name() for encapsulation.</p><p><strong>My Code Practice</strong><br>I practiced by adding books to the library, borrowing them with different member types, testing search_books() to search by title or author, and checking is_available to track available books.</p><p><strong>Challenges I Faced</strong><br>Understanding how __next__() controls iteration and how StopIteration stops the loop was tricky at first. Also implementing different borrowing limits for StudentMember and TeacherMember took me some time to get right.</p><p><strong>How I Solved It</strong><br>I practiced with small examples like a simple CountDown iterator before adding iterators to my main Library project. I also watched tutorials and re-read my class notes on inheritance and polymorphism.</p><p><strong>Conclusion</strong><br>This Library Management System project helped me see how OOP makes code organized, reusable, and closer to real-world systems. It was a great hands-on way to understand classes, inheritance, polymorphism, iterators and scope.</p><figure><img alt="What I see daily as I strive to become better as a self taught developer. The journey is a long one, and that’s great because the journey is the awesome part.Enjoy" src="https://cdn-images-1.medium.com/max/1024/0*6zvCdr289lWblxAk" /><figcaption>Photo by Faith Innocent</figcaption></figure><p><strong>Notebook Link:</strong> <a href="https://drive.google.com/file/d/1SGxG1IFerEGIbvCV1LWgLy2Of0RoUo5P/view?usp=sharing">https://drive.google.com/file/d/1SGxG1IFerEGIbvCV1LWgLy2Of0RoUo5P/view?usp=sharing</a></p><p>#Python #OOP #DataraFlow #Week2 #LibraryManagementSystem</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ea93787966b9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[# My Week 1 Journey into Data Science with Dataraflow]]></title>
            <link>https://medium.com/@faithinnocent774/my-week-1-journey-into-data-science-with-dataraflow-da88d31b6ade?source=rss-5fdc0e3bfa43------2</link>
            <guid isPermaLink="false">https://medium.com/p/da88d31b6ade</guid>
            <dc:creator><![CDATA[Faith  Innocent]]></dc:creator>
            <pubDate>Wed, 09 Sep 2026 11:41:28 GMT</pubDate>
            <atom:updated>2026-09-09T11:41:28.034Z</atom:updated>
            <content:encoded><![CDATA[<p># My Week 1 Journey into Data Science with Dataraflow</p><p>When I first joined Dataraflow, I was scared. I didn’t know Python, I didn’t know what &quot;data science&quot; really meant, and installing software felt like rocket science.</p><p>## What I learned this week:<br>Step 1 was installing Python, VS Code, GitHub Desktop, and Chrome. I learned that all data scientists start with tools. <br>Step 2 was watching the overview videos. I learned that data science is about asking questions and finding answers in data. It’s not just coding, it’s problem solving. <br>Step 3 was creating my first notebook in Google Colab. Writing `print(&quot;Hello Dataraflow&quot;)` and seeing it work made me so happy.</p><p>## Challenges I faced:<br>The hardest part was installing everything without errors. My laptop was slow and I was confused about GitHub. I also struggled to start writing because I thought I had to sound like an expert.</p><p>## How this changed me:<br>This week taught me that you don’t have to know everything to start. You just have to start. Data science isn’t magic, it’s learning one step at a time. I’m excited to see what Week 2 brings.</p><p>Thank you Dataraflow and mentor &lt;669&gt; for the support.</p><p>#Dataraflow #DataScience #Week1 #LearningToCode</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=da88d31b6ade" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>